Rohan Nayak@RohanNayak2
Why LLMs can't write well and how Sherpa fixes it
Why LLMs can't write well and how Sherpa fixes it
- post
- Saved
Written by AI
- TLDR
- Pocket FM's Sherpa combines hierarchical planning, structured story memory, specialized prose models, and critics, trained against writer edits and audience behavior.
- Best for
- Teams building long-form serialized fiction with AI
- The tip
- Separate planning, persistent story state, prose generation, and critique instead of relying on one general-purpose prompt.
- What you need
- A story plan
- Structured canon
- Writer and audience feedback
- Start here
- Plan arcs, episodes, and scenes
- Track character knowledge and unresolved setups
- Draft with critics for continuity and scene quality
postThe source
Rohan Nayak@RohanNayak2Media saved
- Highlights
Creative writing requires tension, emotion, creation and slow resolution of conflict, character traits that go beyond one-liners, and variations in pace. Users must be led to wrong conclusions while receiving clues that keep them hooked.
Good creative writing optimizes the audience's experience, not just the correctness of the answer.
A Narrative World Model keeps a structured record of what has happened, what each character knows, and which promises the story still owes the audience.
Sherpa keeps story facts, character knowledge, and unresolved promises in structured state.
- Excerpt
As we pivoted to AI-written shows, countless people told us this would kill Pocket FM. After six long months, I almost believed them. But then our ecosystem exploded, and not despite AI writing but because of it. Within <2 years, our AI has helped writers create 161 blockbusters on Pocket, including a $100M IP.
From x.com